c58196b795
The last structural fix for the collision class that has bitten three times. Both containers now clone the repository into their OWN named volume instead of bind-mounting a human's working tree, so an agent's local git config cannot capture a human's commits, a credential helper cannot leak a container-only path onto the host, and a `git add -A` cannot sweep another party's in-flight files. Cloned once at startup and never auto-pulled: pulling under a running agent moves files out from under whatever it is mid-edit, which is the same bug again. Accepted knowingly: Claude Code keys per-project memory off the working directory, so moving off the host path starts that memory empty. The corpus in docs/ is the memory that matters and it travels with the clone. Other changes: * docker/agent -> docker/decoder; the launcher is sylph-decoder. Roles, not "the agent", now that there is more than one. * /reborn is gone -- one repository now, so the port reads HANDOFF from its own checkout rather than through a live read-only mount of someone else's tree. * Canary mounts separately at /canary; it stays a fork tracking upstream. * A shared `sylpheed-exchange` volume at /exchange, with tools/ on PATH so `share` is available in both. * The decoder's credential file gets the .host-copy treatment the port already had -- `credential.helper=store` rewrites by rename-over-target, which is EBUSY on a bind mount and reports a fatal that is not one. * Budget split deliberately: decoder 5 cpu / 6 GB, port 3 / 4, leaving room for the planned Referee. "Half the host" was right when there was one agent. Prompts move to docs/agents/ and are rewritten around the protocol: the oracle is the running game, dynamic RE stays with the decoder, each iteration must attempt to refute one claim of the other, and neither may verify its way out of its own role.
164 lines
8.6 KiB
Bash
Executable File
164 lines
8.6 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
# Bring up the headless display, then hand over to the command.
|
|
#
|
|
# Xvfb and openbox are started HERE, as children of PID 1 (tini), rather than by
|
|
# the toolkit scripts. That is the fix for the long-standing "Xvfb and the
|
|
# emulator die on their own every few minutes" note: nothing owned those
|
|
# processes, so nothing kept them alive, and a run could sit for 300 s in front
|
|
# of a visible MAIN MENU reporting "no main menu" because the display had gone.
|
|
# A container-lifetime display makes that failure mode impossible.
|
|
set -euo pipefail
|
|
|
|
log() { printf '[entrypoint] %s\n' "$*" >&2; }
|
|
|
|
DISPLAY="${DISPLAY:-:98}"
|
|
GEOM="${SCREEN_GEOMETRY:-1280x720x24}"
|
|
export DISPLAY
|
|
|
|
# ── Display ──────────────────────────────────────────────────────────────────
|
|
if ! xdpyinfo -display "$DISPLAY" >/dev/null 2>&1; then
|
|
rm -f "/tmp/.X${DISPLAY#:}-lock" "/tmp/.X11-unix/X${DISPLAY#:}" 2>/dev/null || true
|
|
# GLX and RANDR are both required: Canary's window is GTK+OpenGL even when the
|
|
# graphics backend is Vulkan, and xwininfo-based screen oracles need RANDR.
|
|
Xvfb "$DISPLAY" -screen 0 "$GEOM" -ac -nolisten tcp \
|
|
+extension GLX +extension RANDR >/tmp/xvfb.log 2>&1 &
|
|
for _ in $(seq 1 50); do
|
|
xdpyinfo -display "$DISPLAY" >/dev/null 2>&1 && break
|
|
sleep 0.2
|
|
done
|
|
fi
|
|
if ! xdpyinfo -display "$DISPLAY" >/dev/null 2>&1; then
|
|
log "FATAL: no display on $DISPLAY — see /tmp/xvfb.log"
|
|
exit 1
|
|
fi
|
|
|
|
# A window manager is not cosmetic here: without one the emulator window is
|
|
# never mapped at a known position, and every pixel oracle reads the wrong rows.
|
|
if ! pgrep -x openbox >/dev/null 2>&1; then
|
|
openbox >/tmp/openbox.log 2>&1 &
|
|
sleep 0.5
|
|
fi
|
|
log "display $DISPLAY ready ($GEOM)"
|
|
|
|
# ── Vulkan ───────────────────────────────────────────────────────────────────
|
|
# Prefer the real GPU when /dev/dri was passed through; fall back to lavapipe,
|
|
# which is slow but correct and needs no host device.
|
|
LVP=$(ls /usr/share/vulkan/icd.d/lvp_icd*.json 2>/dev/null | head -1)
|
|
if [ "${SYLPH_VULKAN:-auto}" = "sw" ] || [ ! -e /dev/dri/renderD128 ]; then
|
|
[ -n "$LVP" ] && export VK_ICD_FILENAMES="$LVP"
|
|
else
|
|
unset LIBGL_ALWAYS_SOFTWARE
|
|
fi
|
|
# Report what Vulkan ACTUALLY enumerated, not what we asked for. Announcing
|
|
# "host GPU via /dev/dri" because the device node exists is how an agent ends up
|
|
# believing it has hardware while running llvmpipe — an NVIDIA card needs the
|
|
# NVIDIA Container Toolkit, and /dev/dri alone does nothing for it.
|
|
if command -v vulkaninfo >/dev/null 2>&1; then
|
|
vkdev=$(vulkaninfo --summary 2>/dev/null | awk -F= '/deviceName/{gsub(/^ +| +$/,"",$2); print $2; exit}')
|
|
case "${vkdev:-}" in
|
|
"") log "Vulkan: NO DEVICE — vulkaninfo enumerated nothing" ;;
|
|
llvmpipe*|lavapipe*) log "Vulkan: $vkdev (SOFTWARE — correct but slow)" ;;
|
|
*) log "Vulkan: $vkdev (hardware)" ;;
|
|
esac
|
|
fi
|
|
|
|
# ── Build parallelism ────────────────────────────────────────────────────────
|
|
# Bounded by MEMORY, not just cores. A full-parallel build of this tree has
|
|
# OOM-killed the host outright, and inside a half-the-box container the ceiling
|
|
# is lower still. ~1.5 GiB per C++ TU is the rule of thumb that has held.
|
|
# nproc reports the HOST's core count: --cpus is a CFS quota, not a mask. Using
|
|
# it would oversubscribe a half-the-box container by exactly 2x, so read the
|
|
# quota the cgroup actually grants.
|
|
cpus=$(nproc)
|
|
if [ -r /sys/fs/cgroup/cpu.max ]; then
|
|
read -r _q _p < /sys/fs/cgroup/cpu.max || true
|
|
if [ "${_q:-max}" != max ] && [ "${_p:-0}" -gt 0 ]; then
|
|
cpus=$(( (_q + _p - 1) / _p ))
|
|
[ "$cpus" -lt 1 ] && cpus=1
|
|
fi
|
|
fi
|
|
mem_gib=$(awk '/MemAvailable/{printf "%d", $2/1048576}' /proc/meminfo)
|
|
# MemAvailable is the HOST's too under cgroup v2; prefer the container's cap.
|
|
if [ -r /sys/fs/cgroup/memory.max ]; then
|
|
_m=$(cat /sys/fs/cgroup/memory.max)
|
|
[ "$_m" != max ] && mem_gib=$(( _m / 1073741824 ))
|
|
fi
|
|
[ "${mem_gib:-0}" -lt 1 ] && mem_gib=1
|
|
by_mem=$(( mem_gib * 2 / 3 ))
|
|
[ "$by_mem" -lt 1 ] && by_mem=1
|
|
jobs=$(( cpus < by_mem ? cpus : by_mem ))
|
|
export SYLPH_JOBS="$jobs" CARGO_BUILD_JOBS="$jobs" CMAKE_BUILD_PARALLEL_LEVEL="$jobs"
|
|
log "build parallelism: $jobs (cpus=$cpus, mem=${mem_gib}GiB avail)"
|
|
|
|
mkdir -p "$HOME/shots" "$HOME/logs"
|
|
|
|
# ── The repository, cloned into THIS AGENT'S OWN volume ─────────────────────
|
|
# Not a bind mount of a human's working tree. That arrangement bit this project
|
|
# three times: an agent's `git config --local` captured a human's commits, a
|
|
# credential helper leaked a container-only path onto the host, and a `git add
|
|
# -A` swept an agent's in-flight files into somebody else's commit. Separate
|
|
# checkouts make all three impossible rather than discouraged.
|
|
#
|
|
# Cloned ONCE. Never auto-pulled: pulling under a running agent moves files out
|
|
# from under whatever it is mid-edit, which is the same class of bug again.
|
|
if [ ! -d /work/.git ]; then
|
|
echo "[entrypoint] cloning ${SYLPH_REPO_URL:-https://git.mc02.dev/fabi/Sylpheed.git} into /work"
|
|
git clone --quiet "${SYLPH_REPO_URL:-https://git.mc02.dev/fabi/Sylpheed.git}" /work || {
|
|
echo "[entrypoint] clone FAILED -- the agent has no repository" >&2; }
|
|
else
|
|
echo "[entrypoint] /work is at $(git -C /work rev-parse --short HEAD 2>/dev/null) on $(git -C /work rev-parse --abbrev-ref HEAD 2>/dev/null)"
|
|
fi
|
|
|
|
# The shared exchange, for transient files that must not enter git history.
|
|
mkdir -p /exchange/files 2>/dev/null || true
|
|
|
|
# ── Claude Code config ───────────────────────────────────────────────────────
|
|
# Seed ~/.claude.json from the host's read-only copy, then stamp onboarding as
|
|
# complete. Claude Code re-runs its first-run wizard whenever
|
|
# lastOnboardingVersion differs from the installed version — so a container with
|
|
# a newer Claude than the host stops on the theme picker, with no error and no
|
|
# log line, and an unattended agent sits there forever.
|
|
if [ -f "$HOME/.claude.host.json" ] && [ ! -s "$HOME/.claude.json" ]; then
|
|
cp "$HOME/.claude.host.json" "$HOME/.claude.json" 2>/dev/null || true
|
|
fi
|
|
# `credential.helper=store` rewrites this file by rename-over-target, which
|
|
# fails with EBUSY on a bind mount -- reported as `fatal: unable to write
|
|
# credential store`, while the push itself succeeds. A fatal line that is
|
|
# routinely wrong teaches the reader to ignore the one that is real.
|
|
if [ -f "$HOME/.git-credentials.host" ]; then
|
|
cp "$HOME/.git-credentials.host" "$HOME/.git-credentials" 2>/dev/null || true
|
|
chmod 600 "$HOME/.git-credentials" 2>/dev/null || true
|
|
fi
|
|
|
|
CLAUDE_VER=$(claude --version 2>/dev/null | grep -oE '^[0-9][0-9.]*' || echo 0.0.0)
|
|
python3 /usr/local/bin/seed-claude-config.py "$HOME/.claude.json" "$CLAUDE_VER" \
|
|
"$PWD" "${PROJECT_DIR:-/work}" "$HOME" || true
|
|
chmod 600 "$HOME/.claude.json" 2>/dev/null || true
|
|
|
|
# ── Claude Code ──────────────────────────────────────────────────────────────
|
|
if [ "${SYLPH_AUTONOMOUS:-0}" = "1" ]; then
|
|
# Drop the image's default CMD first, or `claude` is handed the literal string
|
|
# "bash" as its prompt and answers a question nobody asked.
|
|
if [ "$#" -eq 1 ] && [ "$1" = "bash" ]; then
|
|
set --
|
|
fi
|
|
# The flag the user asked for. It is refused under root, which is why this
|
|
# image runs as `agent`.
|
|
# Remote Control registers the session with your account so you can chat with
|
|
# the agent from claude.ai or your phone — the point of a detached run being
|
|
# that you are not sitting in front of it. The name is passed EXPLICITLY: the
|
|
# flag's value is optional, so a bare `--remote-control` would swallow the
|
|
# /loop prompt that follows as the session name.
|
|
if [ "${SYLPH_REMOTE:-1}" != "0" ]; then
|
|
set -- --remote-control "${SYLPH_REMOTE_NAME:-sylpheed-agent}" "$@"
|
|
log "Remote Control enabled as '${SYLPH_REMOTE_NAME:-sylpheed-agent}'"
|
|
fi
|
|
# claude-autonomous wraps `claude --dangerously-skip-permissions` in a pty and
|
|
# answers the one-time first-run gates. The Bypass Permissions disclaimer in
|
|
# particular has no config key that skips it, so unattended it hangs forever.
|
|
set -- claude-autonomous "$@"
|
|
log "starting Claude Code with --dangerously-skip-permissions in $(pwd)"
|
|
fi
|
|
|
|
exec "$@"
|